2009/3/17  <[email protected]>:

> a.      Is this the right thing to do?
> 3.      Is there an easier way to track a SOAP request across multiple calls?

I've built a system with a very similar topology to this without using
WS-Addressing: http://funcnet.eu/

When the front-end server receives a request, it generates a unique ID
for it and returns that to the user immediately. It also creates a
database record for the session which contains the job ID, user's
email address and some metadata.

It then calls each of the prediction services synchronously but in
parallel, using a separate thread for each one. As each one returns,
the front-end stores the results in the database, and when the last
one returns, it sets a 'COMPLETE' flag in the database.

The user gets MonitorJob (get current status), CancelJob and
RetrieveScores operations -- obviously the latter results in an error
unless the job is COMPLETE.

Whether this is easier than using WS-A or not, I dunno :-)

Andrew.

-- 
:: http://biotext.org.uk/ ::

Reply via email to